Chat — a council of models

What it makes

Chat is a council rather than a single assistant. Ask once and several models answer in columns you can read side by side, with their reasoning visible where the model exposes it. A data fuser then writes one combined answer out of the columns you kept.

For the questions where one model's answer is not enough — a judgement call, a translation, a piece of code you want a second opinion on before you trust it.

How a session works

1. Pick one model, or a few, from the catalogue of local and cloud models and ask your question.
2. Read the columns beside each other, turn on Think mode where you want the reasoning shown, and drop the answers that miss.
3. Fuse the rest into one answer, or carry the thread on with the model that did best.

What it costs

Chat is included from the Free plan and the free-to-serve models cost no credits at all. Premium models spend credits per answer, and comparing two of them spends both.

Generation spends credits: paid plans include a monthly allowance, the Free plan starts with a one-time welcome grant, and you can top up at any time.

Included from the Free plan.

Questions

Is Chat free to use?

Yes. It opens on the Free plan, and the free-to-serve models answer without spending credits. Premium models spend them — from your plan's monthly allowance, or from the Free plan's welcome grant.

What does comparing models actually do?

It sends the same prompt to each model you picked and lays the answers out in columns. You read them together instead of asking the same thing three times and trying to remember the first answer.

What is the data fuser?

A pass that reads the columns you kept and writes a single answer out of them, so a comparison ends with one thing to take away rather than three things to reconcile.

Can I send images?

Yes. Images go into the conversation as attachments and come back out of it as results, and the models that can see pictures read the ones you send.